Snowflake Intelligence has just gone GA! This blog will break down everything you need to know about this powerful new feature in Snowflake, including how it works and how it can make a significant impact in your organization.
What is Snowflake Intelligence?
Snowflake Intelligence is an AI-powered chat interface within Snowflake that allows users to ask questions, gain insights from their data, and even perform actions directly.
Snowflake Intelligence is powered by Cortex Agents, which serve as the application’s brains. Cortex Agents process the user’s prompt to understand what tools it needs to invoke to retrieve data or execute logic to derive an accurate answer. Cortex Agents can use structured data in Snowflake through Cortex Analyst, unstructured data through Cortex Search, custom logic through tools, and even third-party data through Cortex Knowledge Extensions (CKEs).
And everything is managed in Snowflake’s secure boundary with the standard RBAC and Observability features you’ve come to expect.
How Snowflake Intelligence Works
Upon logging into Snowflake Intelligence, users will be welcomed with a simple chat interface where they can select which agent they want to interact with. The agents available to a given user are all managed through Snowflake’s RBAC policy, meaning you can ensure only the right users have access to specific agents.
Users will submit their prompts in the textbox shown above, which will be directly passed to the selected Cortex Agent. The agent will reason using a Large Language Model (LLM) from leading providers such as OpenAI and Anthropic to decide which tools need to be invoked. The agent will then call the various tools needed to complete the task at hand, and then provide back the answer to the user.
Responses in Snowflake Intelligence can go beyond just simple text. The agent can understand when a response would fit well with a chart and then render that chart directly in Snowflake Intelligence, all without needing to write any custom code.
As mentioned prior, agents can also perform actions such as sending emails, invoking a machine learning model, or searching the web for additional information to supplement the result. The composable nature of Cortex Agents means that you can dream up a wide variety of applications that are tailored to your business.
Snowflake Intelligence Use Cases
Now that we’ve walked through what Snowflake Intelligence is and how it works, let’s highlight a few different use cases on how it can benefit your organization.
Marketing Campaign Analysis
Many retail organizations have a marketing data mart where historical campaigns and promotions data are stored for analytical purposes. Combining this data with actual sales and additional tools can put significant analytical power into the hands of business users without the need for bespoke report development.
The diagram above depicts an example of how one such agent could be built. It leverages Cortex Analyst to retrieve structured data from both the marketing and sales data marts, and a custom tool to invoke a machine learning (ML) endpoint. The ML endpoint would help answer “why” questions about previous marketing campaign performance.
In this case, the model could be a causal inference model designed to estimate the impact of specific promotion tactics, such as the discount amount or channel on overall campaign outcomes.
This solution could help answer questions such as:
What were my 5 worst-performing marketing campaigns last quarter? What made them perform poorly?
How did the campaign “ABC123” perform over the last week?
What’s the most common promotion lever we’ve been using recently? How has it been performing relative to other levers?
Sales / Finance Research Analysis
Another common use case is asking simple analytical questions about your sales data. While that’s possible with Snowflake Intelligence, it can also be taken a step further using Cortex Search and Cortex External Knowledge Sources.
This solution enables you to synthesize insights between your structured sales data, macroeconomic factors via news stories, and internal strategy & market research documents to answer complex questions such as:
Why did revenue dip in Q2 in EMEA?
How could US policy changes affect our sales strategy over the next year?
Have there been any changes to our pricing strategy over the last month?
These are just two examples of the limitless possibilities with Snowflake Intelligence.
How Snowflake’s Broader Capabilities Complement Snowflake Intelligence
Additional features in Snowflake are bringing the vision for Snowflake Intelligence full circle. To get the most out of Snowflake Intelligence, organizations can utilize two important platform ingredients:
A simple, governed way to ingest and organize unstructured content at scale.
A consistent business layer that gives AI agents precise, policy-aware context.
That’s where Snowflake Openflow and Snowflake Semantic capabilities help elevate the Snowflake Intelligence experience. Together, they streamline unstructured ingestion and establish a semantic layer that makes answers from Snowflake Intelligence more accurate, explainable, and auditable while keeping everything in the same secure Snowflake environment.
What is Snowflake Openflow?
Openflow enables the controlled ingestion of unstructured and semi-structured sources, such as emails, PDFs, wikis, call transcripts, and images, directly into Snowflake. Openflow also supports integration with end-user storage and knowledge bases like Microsoft Sharepoint, Google Drive, and Atlassian Confluence, with the ability to replicate role-based access control (RBAC) permissions to Snowflake.
This allows Snowflake Cortex to index and analyze these assets alongside existing structured data with consistent security controls during data movement.
By consolidating data within a single governed environment, teams can apply consistent access policies and streamline retrieval across both structured and unstructured sources.
In practice, this supports integrated agent workflows. For example, combining tabular lookups from Cortex Analyst with document retrieval through Cortex Search. Extending on our sales and financial research use case, organizations can enrich their analyses with market research or news content, improving decision-making without distributing sensitive data across multiple systems.
Semantic Views & Models
Semantic Views & Models provide a unified business layer within Snowflake, defining key metrics, dimensions, and relationships that AI agents can reference when interpreting and generating responses. This structure reduces ambiguity, maintains consistency across teams, and aligns analytical outcomes with governed business definitions.
Because these models operate within Snowflake’s secure framework and honor RBAC policies, organizations can ensure that insights remain traceable and compliant. For example, in our marketing campaign analysis use case, shared definitions for campaign, lift, and conversion enable agents to produce outputs consistent with enterprise reporting standards, while still allowing deeper exploration through custom analytical tools.
Conclusion
AI applications are only as good as the data supporting them. Snowflake’s suite of AI tools has made it easier than ever before to integrate your company’s data with the most powerful AI models available today, ensuring that your next AI application is set up for success.
Start your Snowflake Intelligence pilot today!
Unlock faster, governed AI insights with Snowflake Intelligence. Connect with phData’s experts to design high‑impact agents, wire up Cortex tools, and stand up a secure, RBAC‑aware semantic layer so your teams can ask questions and act with confidence inside Snowflake.
